Gradio app
#9
by
freakynit
- opened
Thank you!
How much GPU do we need for hosting this model?
not much i guess... im using 3090 and its working blazing fast.
Wanted to run it just in vLLM is that possible by using the vLLM serve command ? I don’t see vLLM is mentioned here.
How are you hosting it ? vLLM? Would you mind sharing it ?
Checkout their github repo.. they have full vLLM code given. I was running using transformers.
@firstdears i try to with vllm serve like you getting some package errors if i can solve it sharing it here otherwise i try to run like https://github.com/deepseek-ai/DeepSeek-OCR#vllm-inference in here.
Updated with bounding box support and parsing of raw ocr output: https://github.com/freakynit/deepseek-ocr-gradio 🚀